Satellite remote sensing is an important means to study the infrared radiation characteristics of the earth's atmosphere background. Due to the influence of atmosphere and the limitation of sensor observation conditions, it is impossible to obtain multi-meteorological and multi-detection radiation data. Aiming at this problem, the spectral radiation characteristics of vegetation, water and rock were analyzed based on JHU spectral database. Based on the spectral response of sensors, a model of surface radiation band transformation model based on spectral correlation was established for 3-5 μm. The error was calculated by the stepwise regression method, and the model error was less than 10%. Using MODIS, AIRS and other multi-source remote sensing products, and according to the surface-atmosphere-sensor radiation transfer model, the mid-infrared image of the earth background radiation was simulated. It can simulate the mid-infrared image of the earth background radiation with different temporal and spatial resolution and detection conditions. The results show that the simulation of mid-infrared image with multi-source remote sensing data can achieve large-scale and fine texture image simulation which can be used in remote sensing research.